def train_model(x_train_grid, x_test_grid, y_train_grid, y_test_grid):
"""
train_model function removes features that aren't needed for model training.
:param x_train_grid: x_train dataset for modelling
:param x_test_grid: x_test dataset for modelling
:param y_train_grid: y_train dataset for modelling
:param y_test_grid: y_test dataset for modelling
:return list of trained models with metrics
"""
model_details,param_details,train_model_lists={},{},[]
model_details={
'KNN_Classifier': KNeighborsClassifier(),
'SVM_Classifier': SVC(random_state=42),
'XGB_Classifier': XGBClassifier(random_state=42),
'GaussianNB': GaussianNB(),
'GradientBoosting_Classifier': GradientBoostingClassifier(random_state=42),
'AdaBoost_Classifier': AdaBoostClassifier(random_state=42),
'RandomForest_Classifier': RandomForestClassifier(random_state=42),
'DecisionTree_Classifier': DecisionTreeClassifier(random_state=42),
}
param_details={
'KNN_Classifier': {'n_neighbors': np.arange(1,int(np.sqrt(len(x_train_grid)))).tolist(), 'p': [1,2]},
'SVM_Classifier': {'C':[0.1,1,10,100],'gamma':[1,0.1,0.001] },
'XGB_Classifier': {'learning_rate':[0.05,0.10,0.15],'max_depth':[3,4,5],'min_child_weight':[1,3,5],'gamma':[0.1,0.2,0.3],'colsample_bytree':[0.3,0.4,0.5],'n_estimators':[20,40,50]},
'GaussianNB': {'var_smoothing': np.logspace(0,-9, num=100)},
'GradientBoosting_Classifier': {'learning_rate': [0.05,0.10,0.15],'min_samples_split': np.linspace(0.1,0.5,3),'min_samples_leaf': np.linspace(0.1,0.5,3),'max_depth':[3,5,8],'max_features':['log2','sqrt'],'subsample':[0.5,1.0],'n_estimators':[20,40,50]},
'AdaBoost_Classifier': {'n_estimators': [20,40,50,60],'learning_rate': [0.05,0.10,0.15]},
'RandomForest_Classifier': {'n_estimators': [20,40,50,60],'max_depth' : [4,5,6,7,8],'criterion' :['gini','entropy']},
'DecisionTree_Classifier': {'ccp_alpha': [0.1,.01,.001],'max_depth': [5, 6, 7, 8, 9],'criterion' :['gini', 'entropy']},
}
for key in model_details.keys():
print("Running GridSearchCV for %s." % key)
list_grid_search = GridSearchCV(model_details.get(key), param_details.get(key), cv=5, verbose=1, n_jobs=-1, refit=True)
list_grid_search.fit(x_train_grid,y_train_grid)
predicted_model=list_grid_search.best_estimator_
predicted_model.fit(x_train_grid,y_train_grid)
y_pred = predicted_model.predict(x_test_grid)
y_train_pred = predicted_model.predict(x_train_grid)
train_model_lists.append([key, accuracy_score(y_train_grid,y_train_pred), accuracy_score(y_test_grid, y_pred),
roc_auc_score(y_test_grid,y_pred), precision_score(y_test_grid,y_pred), recall_score(y_test_grid,y_pred), f1_score(y_test_grid,y_pred), predicted_model])
print("GridSearchCV for %s completed." % key)
return train_model_lists
basic_model_df=pd.DataFrame(train_model(X_train, X_test, y_train, y_test),columns=['Model_Name','Train_Accuracy','Test_Accuracy','ROC_AUC','Precision','Recall','F1 Score', 'Model']).sort_values(by='F1 Score', ascending=False)
print('\n######### Model Training using PCA #########\n')
pca_model_df=pd.DataFrame(train_model(X_pca_train, X_pca_test, y_pca_train, y_pca_test),columns=['Model_Name','Train_Accuracy','Test_Accuracy','ROC_AUC','Precision','Recall','F1 Score', 'Model']).sort_values(by='F1 Score', ascending=False)
